Forest fire risk assessment system based on probability grammar model

Abstract

The invention discloses a forest fire risk assessment system based on a probability grammar model. The forest fire risk assessment system comprises a static risk assessment module, a dynamic risk assessment module and a fire risk reasoning module, the static risk assessment module is used for acquiring static risk index data and assessing static risks through a trained first assessment model; the dynamic risk assessment module is used for acquiring dynamic risk index data and assessing a dynamic risk through a trained second assessment model; the fire risk reasoning module is used for constructing a forest model according to the hierarchical relationship between the monitoring points and the areas and the interactive influence relationship between the monitoring points; reasoning the fire risk of each node in the forest model according to each monitoring point; according to the method, the fire reasoning model with the hierarchical relationship is constructed based on the probability grammar module, single-point risk assessment is gradually expanded to overall assessment, and more comprehensive fire risk analysis is provided.

Description

Technical Field

[0001] The invention relates to the field of intelligent control technology, and more specifically to a forest fire risk assessment system based on a probability grammar model. Background Art

[0002] The healthy and sustainable development of forest resources is of great significance to ensuring national timber security, maintaining the balance of the ecosystem and promoting the construction of ecological civilization. However, forest fires, as one of the most frequent natural disasters in the world, seriously threaten the healthy development of forest resources, and thus affect the balance of the ecosystem and the sustainable development of human society.

[0003] The occurrence of forest fires is affected by many factors, including natural conditions, economic and social levels, and human activities. Traditional forest fire risk assessment methods mostly focus on the analysis of meteorological conditions, ignoring the complex coupling relationship and uncertainty among various factors. By building a more scientific and comprehensive forest fire risk assessment system, we can more accurately predict fire risks and improve the efficiency and effectiveness of fire prevention and firefighting.

[0004] At present, common forest fire risk assessment methods include weighted Bayesian network assessment model, fuzzy mathematics method of information diffusion theory, analytic hierarchy process (AHP), entropy weight method, game theory combined weighting method, GIS remote sensing technology and random forest algorithm, etc. These methods perform well in different aspects.

[0005] The weighted Bayesian network assessment model can reflect the dependencies and contribution values ​​among various fire risk factors, and use expert knowledge to build a model with high practical application value. However, this method is highly complex, requires a large amount of data and expertise, has low computational efficiency, and is not suitable for real-time assessment.

[0006] The fuzzy mathematics method of information diffusion theory is suitable for small sample data and can handle situations where sample information is incomplete. It can provide more reliable assessment results, but it has limitations in describing comprehensive risks and needs to be combined with other assessment methods.

[0007] The analytic hierarchy process (AHP) can take into account the subjective judgment of experts by establishing a hierarchical model and calculating the indicator weights based on expert scoring, and is suitable for decision-making problems that are difficult to fully quantify. However, it is highly subjective and is affected by the personal experience and preferences of experts.

[0008] The entropy weight method determines the weight by calculating the information entropy and difference coefficient of each indicator. It is objective and not affected by subjective judgment, but it relies entirely on data and may ignore the importance of expert experience and qualitative analysis. The game theory combined weighting method combines the analytic hierarchy process and the entropy weight method to obtain the comprehensive weight through the optimization method, balancing the advantages of subjective and objective weighting and improving the scientificity and accuracy of the evaluation. However, it may require a large amount of data and complex calculations in practical applications.

[0009] GIS remote sensing technology provides large-scale, high-resolution spatial data to identify and analyze fire risk factors, but it requires professional GIS skills and software, and the data quality depends on remote sensing images. As a machine learning method, the random forest algorithm can process large amounts of data and capture complex nonlinear relationships, but the model has poor interpretability and the training process requires a large amount of data.

[0010] Among the existing evaluation methods, various methods have their own advantages and disadvantages in data accessibility, interpretability, method scalability, and the ability to utilize expert knowledge.

[0011] Therefore, how to achieve a comprehensive fire risk analysis is an issue that technical personnel in this field need to solve urgently. Summary of the invention

[0012] In view of this, the present invention provides a forest fire risk assessment system based on a probabilistic grammar model, which can gradually expand from single-point risk assessment to overall assessment and provide a more comprehensive fire risk analysis.

[0013] In order to achieve the above object, the present invention adopts the following technical solution:

[0014] A forest fire risk assessment system based on a probabilistic grammar model, comprising: a static risk assessment module, a dynamic risk assessment module and a fire risk reasoning module;

[0015] The static risk assessment module is used to obtain static risk indicator data and assess static risk through a trained first assessment model;

[0016] The dynamic risk assessment module is used to obtain dynamic risk indicator data and assess dynamic risk through a trained second assessment model;

[0017] The fire risk reasoning module is used to construct a forest model based on the hierarchical relationship between monitoring points and regions and the interactive influence relationship between monitoring points; and to infer the fire risk of each node in the forest model based on each monitoring point.

[0018] Preferably, the fire risk of each node in the forest model is inferred according to each monitoring point, specifically including: in the forest model, each monitoring point is a leaf node, the upper level of the leaf node is a block node, and the upper level is a regional node;

[0019] Calculate the joint probability based on each monitoring point to determine the block risk;

[0020] The regional risk is determined based on the maximum risk probability value in each block.

[0021] Preferably, it also includes a sorting module and a visualization module; the sorting module is used to sort each area according to the regional risk to obtain low-risk areas and high-risk areas;

[0022] The visualization module is used to present the fire risk assessment results of each area in the form of charts or maps.

[0023] Preferably, it also includes an indicator weight setting module, which is used to configure corresponding weights for each of the static risk indicators and each of the dynamic risk indicators.

[0024] Preferably, the static risk indicators include geographical factors and urban forest characteristics and management and safety inputs.

[0025] Preferably, the geographical factors include: one or more of slope, landform, elevation, normalized difference vegetation index (NDVI), land use type, vegetation type, altitude and slope aspect; the urban forest characteristics include one or more of mountain forest area, land use, distance from roads and distance from water systems; the management and safety investment include one or more of fire prevention publicity and education, forest ranger houses, annual combustible material clearing area, and planned investment in safety funds.

[0026] Preferably, the dynamic risk indicators include environmental and meteorological factors, emergency rescue capabilities and human activity factors.

[0027] Preferably, the environmental and meteorological factors include one or more of precipitation, soil moisture content, net primary productivity of vegetation, temperature, humidity, air pressure, wind speed, wind direction, consecutive rainless days and the most recent precipitation; the emergency rescue capabilities include one or more of intelligent monitoring and fire fighting, number of rescue teams, cross-city coordinated rescue, and rescue equipment; the human activity factors include one or more of distance from roads, distance from settlements, population density, gross national product, and historical fire point density.

[0028] Preferably, the dynamic risk assessment module is also used to update the dynamic risk index in real time.

[0029] Preferably, both the static risk assessment module and the dynamic risk assessment module are provided with multi-source data acquisition ports.

[0030] It can be seen from the above technical solution that compared with the prior art, the present invention discloses a forest fire risk assessment system based on a probabilistic grammar model, which constructs a fire reasoning model with a hierarchical relationship based on the probabilistic grammar module, realizes the gradual expansion from single-point risk assessment to overall assessment, and provides a more comprehensive fire risk analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0032] Figure 1 A schematic diagram of the structure of a forest fire risk assessment system based on a probabilistic grammar model provided by the present invention.

[0033] Figure 2 Schematic diagram of monitoring point distribution in area 1 in an embodiment of the present invention;

[0034] Figure 3 Schematic diagram of monitoring point distribution in area 2 in an embodiment of the present invention;

[0035] Figure 4 This is a schematic diagram of the structure of a forest model constructed according to region 1 in an embodiment of the present invention;

[0036] Figure 5 Schematic diagram of the forest model structure constructed according to region 2 in an embodiment of the present invention. DETAILED DESCRIPTION

[0037] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0038] The embodiment of the present invention discloses a forest fire risk assessment system based on a probabilistic grammar model, comprising a static risk assessment module, a dynamic risk assessment module and a fire risk reasoning module; the static risk assessment module is used to obtain static risk indicator data, and evaluate the static risk through a trained first assessment model; the dynamic risk assessment module is used to obtain dynamic risk indicator data, and evaluate the dynamic risk through a trained second assessment model; the fire risk reasoning module is used to construct a forest model according to the hierarchical relationship between monitoring points and regions and the interactive influence relationship between monitoring points; and the fire risk of each node in the forest model is inferred according to each monitoring point.

[0039] Next, each structure is described in detail:

[0040] In one embodiment, the main function of the static risk assessment module is to process various static data related to forest fires. These data include geographical factors, urban forest characteristics, management and safety inputs, etc. First, the system collects relevant static data through sources such as geographic information systems (GIS), topographic maps, management records and policy documents. These data specifically include geographical factors (such as slope, landform, elevation, normalized vegetation index (NDVI), land use type, vegetation type, altitude, slope aspect, etc.), which are usually obtained through spatial analysis and generally do not need to be updated in real time. In addition, urban forest characteristics (such as mountain forest area, land use, distance from roads, distance from water systems) are also included. These characteristics are obtained through GIS and remote sensing technology, and the update frequency is not high. Management and safety input (such as fire prevention publicity and education, forest protection room, annual cleaning of combustible area, planning and investment in safety funds) come from management records and policy documents, and the latest data is obtained through annual reports or policy updates.

[0041] In order to further implement the above technical solutions, the static risk assessment module obtains multi-source data through the multi-source data receiving port. During the data collection process, the system does not rely solely on a single data source, but makes comprehensive use of multiple data sources to ensure the comprehensiveness and accuracy of the data. For example, geographic factor data can be obtained from the National Geographic Information Center, urban forest feature data can be obtained from the urban planning department, and management and safety input data need to be obtained in coordination with relevant management departments. These data need to be strictly verified and calibrated during the collection process to ensure the accuracy and reliability of the data.

[0042] Furthermore, after data collection is completed, the system cleans, standardizes and normalizes the data. The purpose of data cleaning is to remove noise and outliers in the data and ensure data quality. Standardization is to convert data from different sources to the same scale for subsequent analysis and processing. Normalization is to convert data to the range of [0,100] to eliminate the impact of different dimensions on data processing. The purpose of these steps is to ensure the consistency and availability of data and provide a reliable data basis for subsequent risk assessment.

[0043] For missing data, the system uses a variety of methods to fill in. Common methods include mean filling, median filling, mode filling, and model-based prediction filling. Mean filling is suitable for situations where there are few missing data and the data distribution is relatively uniform; median filling is suitable for situations where the data distribution is somewhat skewed; mode filling is suitable for processing categorical data; model-based prediction filling uses data from other features to make predictions through machine learning models to fill in missing values. The choice of these methods depends on the specific data characteristics and data missingness.

[0044] In order to further implement the above technical solution, a weight setting module is also included, which is used to configure weights for each indicator in the static risk index. Specifically, the system collects expert knowledge through expert survey questionnaires. The system produces questionnaires containing the relationship between various risk assessment indicators and scenario risk levels, and invites experts in related fields to fill them out. The design of these questionnaires needs to fully consider the professional background and experience of different experts to ensure the scientificity and rationality of the questionnaire design. Through questionnaire surveys, the system collects experts' evaluations of the weights of each indicator and the risk levels of each scenario. These expert knowledge provides the system with valuable experience data, which helps to improve the accuracy and reliability of risk assessment.

[0045] Finally, the system uses the processed data for model training and evaluation. The purpose of model training is to build an assessment model that can accurately predict forest fire risks. The input of the model includes the normalized values ​​of various indicators and the weight set provided by experts, and the output is the risk level of each scenario. A lot of data analysis and algorithm optimization are required during the model training process to ensure the accuracy and robustness of the model. Through repeated training and testing of the model, the system finally obtains a risk assessment model that can work effectively in practical applications.

[0046] In one embodiment, the dynamic risk assessment module is mainly responsible for processing various dynamic data related to forest fires, including environmental and meteorological factors, emergency rescue capabilities and human activity factors, etc. Environmental and meteorological factors include precipitation, soil moisture, net primary productivity of vegetation, temperature, humidity, air pressure, wind speed, wind direction, consecutive rainless days, and the most recent precipitation, etc.; the meteorological station can monitor meteorological data such as temperature, humidity, air pressure, wind speed and wind direction, and the soil monitoring station can monitor data such as soil moisture and net primary productivity of vegetation; emergency rescue capabilities include intelligent monitoring and fire fighting, the number of rescue teams, cross-city regional coordinated rescue, and rescue equipment, etc.; human activity factors include distance from roads, distance from residential areas, population density, gross national product, and historical fire point density, etc.

[0047] In order to further implement the above technical solution, the dynamic risk assessment module is also used to update the dynamic risk index in real time. The module collects relevant dynamic data through equipment such as weather stations, soil monitoring stations, emergency management systems and rescue team real-time data.

[0048] Environmental and meteorological factors need to be monitored in real time through equipment such as weather stations and soil monitoring stations, and need to be updated regularly to reflect current environmental conditions;

[0049] The emergency rescue capability is updated through real-time data from the emergency management system and rescue teams to ensure that the assessment model can reflect the current rescue capability. The system obtains real-time rescue data through cooperation with emergency management departments and rescue teams to ensure rapid response and effective disposal when a fire occurs.

[0050] Human activity factors are obtained partly through real-time monitoring and partly through regularly updated statistical data. Population density and gross national product can be obtained through census and economic statistics, while historical fire point density can be obtained through analysis of historical fire records. The system evaluates the impact of human activities on forest fire risk by comprehensively analyzing these data.

[0051] Based on the dynamic characteristics of data, the system updates data regularly or in real time to reflect the current environmental conditions and emergency rescue capabilities. The frequency of dynamic data updates depends on the rate of data change and application requirements. For example, meteorological data needs to be updated in real time, while emergency rescue capability data can be updated regularly as needed. The system ensures the real-time and accuracy of data through an automated data update mechanism.

[0052] Furthermore, after data collection, the system cleans, standardizes and normalizes the real-time monitored data to ensure data quality and consistency. Data cleaning is to remove noise and outliers, standardization is to convert data of different dimensions to the same scale, and normalization is to convert data to the range of [0,100]. Through these steps, the system can ensure the consistency and availability of data and provide a reliable data basis for subsequent risk assessment.

[0053] For missing data, the system uses appropriate methods to fill in the missing data to ensure the integrity and accuracy of the data. Missing value filling methods include mean filling, median filling, mode filling, and model-based prediction filling. The system selects the appropriate filling method based on the specific data characteristics and missing conditions to ensure data integrity.

[0054] In one embodiment, the fire risk reasoning module infers the fire risk of each node in the forest model based on each monitoring point. In the forest model, the leaf node represents each monitoring point, and each monitoring point contains data from the static risk assessment module and the dynamic risk assessment module. The upper level of the leaf node represents the block node, and the upper level is the regional node. Each region is a tree, and the entire forest area includes multiple regions. The probabilistic grammar model first combines the data of the static risk assessment module and the dynamic risk assessment module to obtain the comprehensive assessment data of each monitoring point. These data include static data (such as geographical factors and urban forest characteristics) and dynamic data (such as environmental and meteorological factors, emergency rescue capabilities and human activity factors). The system uses these data as the input of the leaf node, and calculates the risk assessment results of each node layer by layer through the hierarchical structure of the probabilistic grammar model graph.

[0055] The construction of the probabilistic grammar model not only reflects the hierarchical relationship between points and regions, but also describes the mutual influence between different monitoring points. For example, an increase in the fire risk of a certain monitoring point may affect the fire risk of adjacent monitoring points. The system uses the probabilistic grammar model to express this mutual influence relationship and provide more accurate risk assessment results.

[0056] Furthermore, the leaf nodes are the comprehensive risk assessment values ​​of each monitoring point, and the block nodes are in a logical AND relationship. Because to assess the risk status of the entire block, it is necessary to consider the conditions of all monitoring points in the region. The regional nodes are in a logical OR relationship, indicating that as long as there is a block in the region that is in a risk warning state, the region is in a risk warning state. The data structure of the entire forest reflects the geographical relationship between the monitoring points and the regional block divisions.

[0057] Since various elements of a fire may be transmitted to adjacent areas, the leaf nodes of monitoring points under different block nodes will be connected by edges, reflecting the physical causal relationship between the nodes.

[0058] The following uses the physical cause and effect of the physical transfer of temperature to illustrate. The transfer of other physical causes and effects can be calculated similarly. When a fire occurs at a point in the forest, the temperature at the current location is T1. To calculate the temperature change at a distance d, the current temperature at the location d is known to be T2. Since heat mainly causes temperature changes at surrounding points through heat convection and heat radiation, the heat transferred by heat radiation can be calculated using the Stefan-Boltzmann law, q radiation =α*T1 4 / d 2 The heat convection can be calculated using the physical laws of flux q convection =β*(T1-T2). The temperature change is proportional to the total heat flux, so the temperature change is (q radiation +q convection ) / γ, so the temperature at position d is (q radiation +q convection ) / γ+T2, that is (α*T1 4 / d 2 +β*(T1-T2)) / γ. Among them, α, β, and γ are all physical constants.

[0059] In this embodiment, the present invention considers the interaction between each monitoring point, namely physical causality, to calculate the risk probability of the monitoring point corresponding to each leaf node; and further evaluates the risks of block nodes and regional nodes based on each leaf node, thereby realizing risk evaluation from leaf nodes to regional nodes, and from single points to the whole.

[0060] The state of the entire forest area at a certain moment can be represented by a forest structure containing state parameters. As time goes by, multiple forest structures can be formed. Multiple forest structures reflect the changes of the overall state over time. Therefore, the entire data structure contains changes in space, cause and effect, and time.

[0061] When the system is working, the static risk assessment module inputs the normalized static risk factors and the corresponding weights of each factor into the neural network model trained by the corresponding module to obtain the static risk level of each point. Similarly, the dynamic risk assessment module inputs the normalized dynamic risk factors and the corresponding weights of each factor into the neural network model trained by the corresponding module to obtain the dynamic risk level of each point. The static risk level and dynamic risk level of each point are used as inputs to each leaf node of the probability grammar model of the fire risk comprehensive assessment reasoning module. When calculating the fire risk assessment results of each area, the system adopts a bottom-up calculation method.

[0062] During the specific calculation process, when encountering the node, the system calculates according to the total probability formula, that is, considering the joint probability of all child nodes under the node. The total probability formula is used to integrate the risk status of each child node, so as to accurately reflect the overall risk situation of the node. When encountering the or node, the system takes the maximum value of all child nodes under the node, that is, max processing. In this way, the system can identify the overall risk status of the node when any child node is in a risky state, thereby effectively assessing the fire risk in the area.

[0063] After the calculation is completed, the system sorts the risk assessment values ​​of all areas and identifies high-risk areas and low-risk areas. The sorting results can intuitively display the risk level of each area, which is convenient for managers to make decisions and respond. The system presents the fire risk assessment results of each area in the form of charts or maps, such as generating a risk heat map to show the risk distribution of each area, or generating a risk trend map to show the changing trend of the risk level. These charts and maps can help managers quickly identify high-risk areas, formulate corresponding fire prevention measures and response strategies, and help forest rangers better formulate patrol routes.

[0064] After obtaining the probability grammar model at a certain moment, some dynamic risk factors will spread outward due to physical causality. From this, we can obtain the probability grammar model at different moments, that is, the forest structure that changes over time. Especially in the scene of forest fire. When a fire occurs, the fire will spread to the surrounding points. Without considering other factors, it may spread to the surroundings. In the case of wind, it will spread faster in the direction of the wind.

[0065] Next, the present invention is further described with reference to specific cases:

[0066] In this case, risk assessment is performed on Area 1 and Area 2, such as Figure 2 and Figure 3 These two areas are divided by factors such as water systems or firebreaks that can eliminate fire risks. Each area is divided into several blocks, each of which contains several monitoring points. For monitoring points that are close to each other in different blocks, physical factors will spread during a fire. Figure 3 and Figure 4 In the constructed probabilistic grammar model, some leaf nodes are connected.

[0067] 1. Perform static risk assessment on Area 1 and Area 2 as follows:

[0068] The system first collects relevant static data through sources such as geographic information systems (GIS), topographic maps, management records and policy documents, including geographical factors: slope, landform, elevation, normalized vegetation index (NDVI), land use type, vegetation type, altitude, slope aspect, etc. Urban forest characteristics: mountain forest area, land use, distance from roads, distance from water systems. Management and safety investment: fire prevention publicity and education, forest ranger houses, annual clearing of combustible areas, and planning of safety funds. We consider slope, landform, and elevation here. After data collection, data quality and consistency are ensured through data cleaning, standardization and normalization. For missing data, mean filling and model-based predictive filling methods are used. The following results are obtained:

[0069] Point Slope (degrees) Landform(category) Elevation (m) No.10 15 3 100 No.11 18 3 150 No.12 17 3 200 No.13 14 3 230 … … … … No.74 20 1 500

[0070] Then, the system designs a questionnaire containing the relationship between risk assessment indicators and scenario risk levels, and invites experts to fill it out. The experts' evaluation of the weight of each indicator is collected to form an expert weight set.

[0071] slope 0.3 Landforms 0.15 Elevation 0.55

[0072] In data processing, all static risk factor data are normalized to the range of [0,100], and the following results are obtained:

[0073] Point Slope (degrees) Landform(category) Elevation (m) No.10 60 100 20 No.11 72 100 30 No.12 68 100 40 No.13 56 100 47.5 … … … … No.74 80 33.3 100

[0074] These results are then input into the neural network model in the module to obtain the risk level of the corresponding area.

[0075] Block Risk Level 1 2.1 2 4.3 3 7.0 4 5.1 5 3.3 6 7.0 7 5.5

[0076] 2. Perform dynamic risk assessment on Area 1 and Area 2 as follows

[0077] Data on environmental and meteorological factors, emergency rescue capabilities, and human activity factors are collected through equipment such as meteorological stations, soil monitoring stations, and emergency management systems, including environmental and meteorological factors: precipitation, soil moisture content, net primary productivity of vegetation, temperature, humidity, air pressure, wind speed, wind direction, consecutive rainless days, and the most recent precipitation. Emergency rescue capabilities: intelligent monitoring of firefighting, number of rescue teams, cross-city regional coordinated rescue, rescue equipment, etc. Human activity factors: distance from roads, distance from settlements, population density, gross national product, historical fire point density, etc. For a simple example, we consider temperature, humidity, and wind speed here.

[0078]

[0079]

[0080] Likewise, the data was cleaned, standardized, and normalized to ensure data quality and consistency.

[0081] All dynamic risk factor data are normalized to the range of [0,100], and weighted summation is performed based on real-time updated data to obtain the dynamic risk value of each point. The results are as follows:

[0082] These results are then input into the neural network model in the module to obtain the risk level of the corresponding area.

[0083] Point Temperature(℃) humidity(%) Wind speed (m / s) No.10 83.33 25 62.5 No.11 100 0 87.5 No.12 66.67 75 50 No.13 66.67 75 50 … … … … No.74 83.33 25 62.5

[0084] Block Risk Level 1 1.9 2 4.0 3 5.9 4 5.0 5 3.8 6 6.3 7 5.1

[0085] 3. Fire risk reasoning

[0086] The system uses static risk values ​​and dynamic risk values ​​as inputs to each leaf node of the probability grammar model. It reflects the hierarchical relationship and causal relationship of each monitoring point, block, and region by constructing a forest structure of the probability grammar model, connecting the leaf nodes with causal relationships in the forest structure, and multiple forest structures that change dynamically over time. The system calculates the risk assessment results of each node from the bottom up. Since the probability grammar graph in the embodiment has only one layer or structure, it is only necessary to solve the maximum risk level of each block. After the calculation is completed, the system sorts the risk assessment values ​​of all areas, and the results are as follows

[0087]

[0088]

[0089] The system displays the assessment results in the form of charts and maps, helping managers quickly identify high-risk areas and formulate corresponding fire prevention measures and response strategies.

[0090] For dynamic risk factors that change over time and form multiple probabilistic grammar forest structures, the system simulates fire occurrence and spread scenarios, analyzes changes in fire risks at different time points, and provides prediction and early warning support.

[0091] Through the above steps, the system builds a forest fire risk assessment system based on a probabilistic grammar model, which can effectively assess the fire risk of each monitoring point, block and region, and provide intuitive assessment results and early warning support to help managers formulate fire prevention measures and response strategies.

[0092] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0093] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A forest fire risk assessment system based on a probabilistic grammar model, characterized in that: include: Static risk assessment module, dynamic risk assessment module and fire risk reasoning module; The static risk assessment module is used to obtain static risk indicator data and assess static risk through a trained first assessment model; The dynamic risk assessment module is used to obtain dynamic risk indicator data and assess dynamic risk through a trained second assessment model; The fire risk reasoning module is used to construct a forest model based on the hierarchical relationship between monitoring points and regions and the interactive influence relationship between monitoring points; And the fire risk of each node in the forest model is inferred based on each monitoring point.

2. A forest fire risk assessment system based on a probabilistic grammar model according to claim 1, characterized in that: The inference of the fire risk of each node in the forest model according to each monitoring point specifically includes: in the forest model, each monitoring point is a leaf node, the upper level of the leaf node is a block node, and the upper level is a regional node; Calculate the joint probability based on each monitoring point to determine the block risk; The regional risk is determined based on the maximum risk probability value in each block.

3. A forest fire risk assessment system based on a probabilistic grammar model according to claim 2, characterized in that: It also includes a sorting module and a visualization module; the sorting module is used to sort each area according to the regional risk to obtain low-risk areas and high-risk areas; The visualization module is used to present the fire risk assessment results of each area in the form of charts or maps.

4. A forest fire risk assessment system based on a probabilistic grammar model according to claim 1, characterized in that: It also includes an indicator weight setting module, which is used to configure corresponding weights for each of the static risk indicators and each of the dynamic risk indicators.

5. A forest fire risk assessment system based on a probabilistic grammar model according to claim 1, characterized in that: The static risk indicators include geographical factors and urban forest characteristics and management and safety inputs.

6. A forest fire risk assessment system based on a probabilistic grammar model according to claim 5, characterized in that: The geographical factors include: one or more of slope, topography, elevation, normalized difference vegetation index (NDVI), land use type, vegetation type, altitude and slope aspect; the urban forest characteristics include one or more of mountain forest area, land use, distance from roads and distance from water systems; the management and safety investment includes one or more of fire prevention publicity and education, forest ranger houses, annual combustible material clearing area, and planned investment in safety funds.

7. A forest fire risk assessment system based on a probabilistic grammar model according to claim 1, characterized in that: The dynamic risk indicators include environmental and meteorological factors, emergency rescue capabilities and human activity factors.

8. A forest fire risk assessment system based on a probabilistic grammar model according to claim 7, characterized in that: The environmental and meteorological factors include one or more of precipitation, soil moisture content, net primary productivity of vegetation, temperature, humidity, air pressure, wind speed, wind direction, consecutive rainless days and the most recent precipitation; the emergency rescue capabilities include one or more of intelligent monitoring and fire fighting, number of rescue teams, cross-city coordinated rescue, and rescue equipment; the human activity factors include one or more of distance from roads, distance from settlements, population density, gross national product, and historical fire point density.

9. A forest fire risk assessment system based on a probabilistic grammar model according to claim 1 or 8, characterized in that: The dynamic risk assessment module is also used to update the dynamic risk index in real time.

10. A forest fire risk assessment system based on a probabilistic grammar model according to claim 1, characterized in that: The static risk assessment module and the dynamic risk assessment module are both provided with multi-source data acquisition ports.

Images